Earlier this year, ML6 developed the predictive control platform Waizu that is entirely geared towards optimization of industry processes. The platform combines ML6‘s longstanding expertise in machine learning for predictive maintenance with pioneering research in the field of reinforcement learning and Google's latest technological developments in hardware and cloud. This revolutionary software was developed for industry processes and allows companies to optimize numerous processes.
Saroléa and ML6 joined forces to combine their respective know-how in electric powertrains and AI-driven process controls and applied those to the electric powertrain of the MANX7, Saroléa’s latest model. They use Google's Edge TPU, a processor that was specifically developed for artificial intelligence, and installed it on the engine itself.
"To this day, artificial intelligence is made possible by first sending all of a system’s data via the internet to the cloud. Then, based on that, models are initialized (by letting them learn) and the trained models are then sent back to the machine. Thanks to the Edge TPU, the AI (with its 'trained models') can intervene fast and more efficiently. In addition, each individual system can learn more," says Nicolas Deruytter, managing director of ML6, which works closely together with Google Cloud.
"Over the past four years, we have already collected millions of data lines and developed a basic configuration based on this," says Bjorn Robbens, CEO of Saroléa. "This will quickly add up to billions and such an ongoing data feed will keep optimizing the model even more. Moreover, all motorcycles are now connected with each other via the IoT chip which means they learn from each other, so to speak. The more motorcycles and the more data we have, the better the optimization will be. Because of reinforcement learning, each engine will know exactly which settings are optimal in which situation, leading to a better performance and more autonomy."
Efficiency and lifespan
Countless parameters in a vehicle can be mapped and optimized but that would be impossible to do manually. Furthermore, data vary according to every single situation and location. How is power controlled? What temperatures are reached inside the vehicle? How is the battery managed? The chip captures and interprets. Subsequently, the software determines the best settings. This is how Saroléa and ML6 plan to focus on more autonomy, predictive maintenance and an improved battery performance.
"On the one hand, this allows to increase the vehicle’s efficiency, leading to more autonomy as a direct result. On the other hand, we also extend the battery’s lifecycle. The way in which a battery is used and charged helps determine its lifespan. If we can evolve towards a model in which the battery is charged and used in ideal circumstances, its autonomy could improve by 20 to 40% and the battery will last for years," says Bjorn Robbens.
The potential of this technological application is enormous and reaches far beyond the motorcycle world. That is why Saroléa intends to market their technology in the near future and make it available to the entire automotive sector via a platform. "Buses, trucks, bicycles, automated guided vehicles; we are looking at the bigger picture here. We want to package the technology and know-how we have built up with our partner, ML6, and offer an off-the-shelf solution to other OEM players,” concludes Bjorn Robbens.